Using singular perturbated systems of differencial equations of infinite order for countable markov chains analysis Текст научной статьи по специальности «Математика»

УДК 517.937+ 517.928.2+ 519.217.2

Vasilyev S.A., Tzareva G.O.

Peoples' Friendship University of Russia, Moscow, Russia

USING SINGULAR PERTURBATED SYSTEMS OF DIFFERENCIAL EQUATIONS OF INFINITE ORDER FOR COUNTABLE MARKOV CHAINS ANALYSIS

ABSTRACT

Tikhonov-type Cauchy problems are investigated for systems of ordinary differential equations of infinite order with a small parameter u and initial conditions. It is studying the singular perturbated systems of ordinary differential equations of infinite order of Tikhonov-type x = f (x(t, gx ), y(t, gy ), t), ¡У == F(x(t, gx ), y(t, gy ), t) with the initial conditions

x(t0, gx) = gx, y (t0, gy ) = gy, where x, f e X, X e Rn are n-dimensional functions; y, F e Y, Y ^ / are infinite-dimensional functions and t e[t0, tl ] (t0 < tl <ю), t e T, T e R ; gx e X and gy e Y are given vectors, u > 0 is a small real parameter. The results may be applied to the queueing networks, which arise from the modern telecommunications.

KEYWORDS

Systems of differential equations of infinite order; singular perturbated systems of differential equations; small parameter; v countable Markov chains.

Васильев С.А., Царева Г.О.

Российский университет дружбы народов, г. Москва, Россия

ИСПОЛЬЗОВАНИЕ СИНГУЛЯРНО ВОЗМУЩЕННЫХ СИСТЕМ ДИФФЕРЕНЦИАЛЬНЫХ УРАВНЕНИЙ ДЛЯ АНАЛИЗА СЧЕТНЫХ МАРКОВСКИХ

ЦЕПЕЙ

АННОТАЦИЯ

В статье исследованы задачи Коши для систем обыкновенных дифференциальных уравнений бесконечного порядка с малым параметром u тихоновского-типа

x = f (x(t, gx ), y(t, gy ), t), uy == F(x(t, gx ), y(t, gy ), t) с начальными условиями x(t0, gx) = gx, y(t0, gy) = gy, гдеx, f e X, X e Rn - функции конечного числа измерений; y, F e Y, Y ^ / - функции бесконечного числа измерений и t e[t0, tj ] ( t0< tj <ю), t e T, T e R ; gx e X, а также gy e Y заданные векторы, u > 0 - малый

параметр. Результаты данной работы могут быть применены для анализа прикладных задачах в теории массового обслуживания.

КЛЮЧЕВЫЕ СЛОВА

Системы дифференциальных уравнений бесконечного порядка; сингулярно возмущенных системы дифференциальных уравнений; малый параметр; счетные цепи Маркова.

Introduction

The recent research of service networks with complex routing discipline in [16], [17], [18] transport networks [1], [4], [5] faced with the problem of proving the global convergence of the solutions of certain infinite systems of ordinary differential equations to a time-independent solution. Scattered results of these studies, however, allow a common approach to their justification. This approach will be expounded here. In work [11] the countable systems of differential equations with bounded Jacobi operators are

studied and the sufficient conditions of global stability and global asymptotic stability are obtained. In [10] it was considered finite closed Jackson networks with N first come, first serve nodes and M customers. In the limit M ^ œ, N ^ œ, M / N ^ A > 0, it was got conditions when mean queue lengths are uniformly bounded and when there exists a node where the mean queue length tends to infty under the above limit (condensation phenomena, traffic jams), in terms of the limit distribution of the relative utilizations of the nodes. It was deriven asymptotics of the partition function and of correlation functions.

Cauchy problems for the systems of ordinary differential equations of infinite order was investigated A.N. Tihonov [13], K.P. Persidsky [12], O.A. Zhautykov [19], [20], Ju. Korobeinik [7] other researchers.

It was studied the singular perturbated systems of ordinary differential equations by A.N. Tihonov [14], A.B. Vasil'eva [15], S.A. Lomov [9] other researchers.

A particular our interest is the synthesis all these methods and its applications in telecommunications. In this paper we apply methods from [11] for the singular perturbated systems of ordinary differential equations of infinite order of Tikhonov-type.

TIKHONOV-TYPE CAUCHY PROBLEMS FOR SYSTEMS OF ORDINARY DIFFERENTIAL EQUATIONS OF INFINITE ORDER WITH A SMALL PARAMETER

Let us consider Tikhonov-type Cauchy problems for systems of ordinary differential equations of infinite order with a small parameter u and initial conditions:

* = f( *(t, gx X y(t, g y X1X

¡y = F (x(t, gx ), y(t, gy ), t ); (1)

X(t0 , gx ) = gx,

y (t0. gy ) = gy ,

where x, f e X, X e Rn are n-dimensional functions; y, F e Y, Y ^ lx are infinite-dimensional functions and t e [t0, tl ] (t0< tl < ro), t e T, T e R ; gx e X and gy e Y are given vectors, ¡u> 0 is a small real parameter; x(t, gx) and y(t, gy) are solutions of (1)-(2). Given functions f (x(t, gx ), y(t, gy ), t) and F(x(t, gx ), y (t, gy ), t) are continuous functions for all variables. Let S is an integral manifold of the system (1)-(2) in X x Y x T . If any point t e[t0, t1 ] (x(t *), y(t*), t*) e S of trajectory of this system has at least one common point on S this trajectory (x(t, G), y(t, g), t) e S belongs the integral manifold S totally. If we assume in (1)-(2) that u = 0 than we have a degenerate system of the ordinary differential equations and a problem of singular perturbations

x = f (x(t, gx), y (t), t),

0 = F (x(t, gx ), y(t), t); (3)

x(t0, gx ) = gx,

where the dimension of this system is less than the dimension of the system (1)-(2), since the relations F(x(t), y(t), t) = 0 in the system (3) are the algebraic equations (not differential equations). Thus for the system (3) we can use limited number of the initial conditions then for system (1)-(2). Most natural for this case we can use the initial conditions x(t0,gx) = gx for the system (3) and the initial conditions

y (t0, gy ) = gy disregard otherwise we get the overdefined system. We can solve the system (3) if the equation F(x(t), y(t), t) = 0 could be solved. If it is possible to solve we can find a finite set or countable set of the roots yq (t, gx ) = uq (x(t, gx ), t) where q e N.

If the implicit function F(x(t), y(t), t) = 0 has not simple structure we must investigate the question about the choice of roots. Hence we can use the roots yq (t, gx ) = uq (x(t, gx ), t) (q e N) in (3) and solve the degenerate system

= f (xd (t, gx I Uq (xd (t, gx I t1 t); (4)

(t0, gx ) = gx .

Since it is not assumed that the roots yq (t, gx ) = uq (x(t, gx ), t) satisfy the initial conditions of the Cauchy problem (1)-(2) (yq (t0) ^ gx, q e N), the solutions y(t, gy ) (1)-(2) and yq (t, gx) do not close to each other at the initial moments of time t >0 . Also there is a very interesting question about behaviors of the solutions x(t, gx) of the singular perturbated problem (1)-(2) and the solutions xd (t, gx) of the degenerate problem (4). When t = 0 we have x(t0, gx) = xd (t0, gx) . Do these solutions close to each other when t e (t0, tx ] ? The answer to this question depends on using roots yq (t, gx) = uq (x(t, gx), t) and the initial conditions which we apply for the systems (1)-(2) and (3).

LOCAL EXISTENCE THEOREM FOR CAUCHY PROBLEMS FOR SYSTEMS OF ORDINARY DIFFERENTIAL EQUATIONS OF INFINITE ORDER

Let Tikhonov-type Cauchy problems for systems of ordinary differential equations of infinite order with a small parameter ¡u> 0 and initial conditions (1)-(2) has a form:

z = P( z (t, G, u), t, u); z (t0, G, u) = G,

z = (x^ xn, y^ y2,...)T,

P(z (t, G, u), t, u) = f f2,..., fn, uF u F2,..)T, ()

G (gxl, gx2, ..., gxn , gyl, gy2,...)

where P( z (t, G, u), t, u) is the infinite-dimensional function; G is the given vector; t e^t0, tl ] ( t0< t1 ).

Let z(t, G, u) be a continuously differentiable solution of the Cauchy problems (5) then there are <^(t, G, u) = dz(t, G, u)/ dG, W(t, G, u) = dz(t, G, u)/ du where <^(t, G, u) and W(t, G, u) satisfy of the system of ordinary differential equations in variations:

z = P(z(t, G, u), t, u), 6(t, G, u) = Jz (t, G, u)®(t, G, u),

W(t, G, u) = Jz (t, G, u)W(t, G, u) + Au (t, G, u); (6)

z(t0, G, u) = G, 6(t0, G, u) = I,W(t0, G, u) = 0,t0 e T,

where Jz (t, G, u) = (SPt / dz}.=1 is Jacobi matrix, I is an identity operator and

Au (t, G, u) = (dP, / du)?=l is a vector.

Theorem 1 (local existence theorem). Let P(z (t, G, u), t, u), Jz (t, G, u), Au(t, G, u) be

continuous and meet Gelder's local condition with z e US(G) then the system (6) has only one solution, which

meet the conditions z(t0, G, u) = G, z(t, G, u) eUs (G). Thus z(t, G, u) continuously differentiable with respect to the initial condition, and its derivative meet the equation (6).

Proof. This statement is following from [3] (theorem 3.4.4) when the unlimited operator be A = 0. End proof.

The behavior of the solution z(t, G, u) (5) and the nonnegative condition for the off-diagonal elements of the matrix Jz (t, G, u) is demonstrated by the following theorem.

Theorem 2. Let the solution z of (5) be z (t, G, u) e lx for any t > 0, G e lx and u . The following claims are equal: (i) the off-diagonal elements Jz (t, G, u) are non-negative for any G; (ii) for any G and any vector h e lx, h > 0, z(t, G + h, u) > z(t, G, u) .

Proof. Let us examine a convex set Z , and z(t, G, u) e Z for any G e Z, derivative 6(t, G, u) of function z(t, G, u) can be specify by simultaneous equations (6). In that case the following formula is

fair for any G0, G1 e Z :

z (t, G\ u) - z(t, G u) =

d , (7)

= fo(t,y(s),u)(Gl -G0)ds

where y(s) = (1 - s)G0 + sG\0 < s < 1.

In fact the function z(t,G, u) transfer the segment y(s) into the curve z(t,y(s), u) in (7). The following formula is fair because of the continuous differentiability of function z(t, G, u)

z (t u)=z(t, G u)^m» ds

J0 ds

By the formula of complex derivative

d- ('fsX u) = dG (r(. s))f( s).

ds dG

Recalling that dz / dG = O and y'(s) = G1 - G0, with z =1 we get (7). Let us suppose that statement (i) is fair. So because of (7)

z(t, G + h, u) - z(t, G, u) = f O(t, y(s), u)hds,

where y(s) = G + sh, 0 < s < 1. Because of non-negativeness of function Jz (t, G, u) outside of diagonal from (7) we get O(t, y(s), u) ^ 0, so O(t, y(s), u)h > 0 whence we get statement (ii).

Let us suppose that (ii) is fair. Under the conditions of Theorem 1 P, Jz with z e US(G) be continuous and meet Gelder's local condition. Let Gelder's local condition be P P P< M0P J P< M1, and there are numbers 8 >0,5 = min(s / M0, 1 / M1). Let z(t, G, u) = G + z*(t, G, u) be a solution of (7), where z (t, G, u) is a fixed point of Picard's mapping ) = [ P(G + 6(z))dT under conditions

t e [t0 - 81, t0 + 81],81 < 8 . Mapping ^^ is contraction with coefficient X = 81M1 < 1. Consider the approximation to solution z(t, G, u) = G + z (t, G, u) = G + (t -10)P(z(t, G, u), t, u) . We can see that

Pz(t, G, u) - z(t, G, u)P= =P z* (t, G, u) - z* (t, G, u)P<

Uz (t, G, u) - z(t, G, u)P,

1 - X

Uz(t, G, u) - z(t, G, u) =

= f 'P(G + (z -10)P)dz - ['Pdz =

= f (P(G + (z-10)P) - P)dz = D.

Because of the derivative of the function P is limited and P meet Gelder's local condition with the constant Mx, where P P(G + (z - t0)P(G)) - P(G)P< Mx P(z - t0)P(G)P< M0 Mx \z -10 |, so

P D P< M0 M1(t - t0f /2(1 - X), or P z (t, G, u) - z(t, G, u)P< M0 M1(t -10)2 / 2(1 - X). Using this estimation and for all small Q > 0 we have that

0 < z(t, G + Qej, u) - z(t, G, u) = Ce} + (t - P(G + ) -P(G)] + y(G, t), where Py(G, t)P< M0M1(t -10)2 / 2(1 - X) and ej

is a vector, which has all coordinates equal to 0 but j -th coordinate equal to 1. Component i j of this inequality is given by 0 < (t -10)[P (G + Qej) - P (G)] + yt (G, t) . Dividing by t -10>0 and directing t ^ t0 on the right, considering yi (G, t)/ (t -10) ^ 0 we get 0 < P(G + Qe.) - P(G) . Let us divide last expression by Q

and direct Q ^ 0

0 < PG+jm=IP=Jj,

Q dGt j

what is mean the fairing of statement (i). End proof.

Theorem 3. Let 6 be Markovian mapping and G0, G1 e X, t > 0, u>0 than

P z(t, G\ u) - z(t, G0, u)P<PG' - G0 P.

Proof. Using (6) from the proofing of theorem 4 we have

Pz(t,G\u)-z(t,Gu)P< fp6(t,y(s))(Gl -G0)Pds. (8)

J 0

Let function 6(t,/(s)) is Markovian mapping for any

Pt > 0,s e [0,1],P6(t,r(s))(G1 -G°)P<PG1 -G°P.

Estimating the integral, considering this inequality, we get required. End proof.

This theorem shows us the following sufficient condition for the boundedness of the norm-solution

z(t, G, u).

Corollary fact from theorem 3. Let 3G* e X : z(t, G*, u) = G*. Then Pz(t, G, u) - G* <PG - G*P with t > 0, G e X.

This fact we can use for solutions analysis of the systems (5). Conclusions

The boundaries of applications and possible generalizations. Some works in the routing disciplines. All systems can be analyzed for the global stability but with some condition that the convergence to the steady-state solution will not coordinate-wise, but the norm. We have seen that the most serious constraints of our methods are non-negativity of the Jacobi matrix off-diagonal elements and the availability of the first integral, which equal to the sum of the components. It would be interesting to understand the physical meaning of these conditions. It is necessary to remember that such systems describes the behavior of the queue lengths on the devices. Roughly speaking, zk is the proportion of units in the queue for a service, to which there is at least k requests (including requests, which are serviced at the moment). Non-negative elements of the Jacobi matrix indicate that the rate of change of zk (i.e., the time derivative of zk) can only grow at the expense of zj with j ^ k. It can be reduced (or decrease) only due to uk. Thus, with the increase of the portion of queues with a minimum number of requests j in the system, the percentage change in intensity with the minimum number of queues requests k ^ j can only increase.

References

1. Afanassieva L.G., Fayolle G., Popov S. Yu. Models for Transportation Networks. J. Math. Science. Vol.84, Issue 3, 1997, pp. 1092-1103.

2. Daletsky Y.L., Krein M.G. Stability of solutions of differential equations in Banach space. Moscow, Science Pub., 1970.

3. Henry D. Geometric theory of semilinear parabolic equations. Lecture Notes in Mathematics. Springer-Verlag, Berlin, 1981.

4. Khmelev D. V., Oseledets V.I. Mean-field approximation for stochastic transportation network and stability of dynamical system: Preprint No. 434 of University of Bremen, 1999.

5. Khmelev D. V. Limit theorems for nonsymmetric transportation networks. Fundamentalnaya i Priklladnaya Matematika. Vol. 7, no. 4, 2001, pp. 1259 - 1266.

6. Kirstein B. M., Franken D. E., Stoian D. Comparability and monotonicity of Markov processes. Theory of probability and its applications.Vol. 22, Issue 1, 1977, pp.43-54.

7. Korobeinik Ju. Differential equations of infinite order and infinite systems of differential equations. Izv. Akad. Nauk SSSR Ser. Mat. Vol. 34, 1970, pp. 881 - 922.

8. Krasnoselsky M.A., Zabreyko P.P. Geometrical methods of nonlinear analysis. Springer-Verlag, Berlin, 1984.

9. Lomov S. A. The construction of asymptotic solutions of certain problems with parameters. Izv. Akad. Nauk SSSR Ser. Mat. Vol. 32, 1968, pp. 884 - 913.

10. Malyshev V. and Yakovlev A. Condensation in large closed Jackson networks. Ann. Appl. Probab. Vol. 6, no. 1, 1996, pp. 92115.

11. Oseledets V. I., Khmelev D. V. Global stability of infinite systems of nonlinear differential equations, and nonhomogeneous countable Markov chains. Problemy Peredachi Informatsii (Russian), Vol. 36 , Issue 1, 2000, pp. 60-76.

12. Persidsky K.P. Izv. AN KazSSR, Ser. Mat. Mach., Issue 2, 1948, pp. 3-34.

13. Tihonov A. N. Uber unendliche Systeme von Differentialgleichungen. Rec. Math. Vol. 41, Issue 4, 1934, pp. 551-555.

14. Tihonov A. N. Systems of differential equations containing small parameters in the derivatives. Mat. Sbornik N. S. Vol. 31, Issue 73, 1952, pp. 575 - 586.

15. Vasil'eva A. B. Asymptotic behaviour of solutions of certain problems for ordinary non-linear differential equations with a small parameter multiplying the highest derivatives. Uspehi Mat. Nauk. Vol. 18, Issie 111, no. 3 , 1963, pp. 15 - 86.

16. Vvedenskaya N.D., Dobrushin R.L., Kharpelevich F.I. Queueing system with a choice of the lesser of two queues - the asymptotic approach. Probl. inform. Vol. 32, Issue 1,1996, pp.15-27.

17. Vvedenskaya N.D., Suhov Yu.M. Dobrushin's Mean-Field Approximation for a Queue with Dynamic Routing. Markov Processes and Related Fields. Issue 3, 1997, pp. 493-526.

18. Vvedenskaya N.D. A large queueing system with message transmission along several routes. Problemy Peredachi Informatsii. Vol. 34, no. 2, 1998, pp. 98-108.

19. Zhautykov O. A. On a countable system of differential equations with variable parameters. Mat. Sb. (N.S.). Vol. 49, Issue 91, 1959, pp. 317 - 330.

20. Zhautykov O. A. Extension of the Hamilton-Jacobi theorems to an infinite canonical system of equations. Mat. Sb. (N.S.). Vol. 53, Issue 95, 1961, pp. 313 - 328.

Поступила 21.10.2016

Об авторах:

Васильев Сергей Анатольевич, доцент кафедры прикладной информатики и теории вероятностей Российского университета дружбы народов, кандидат физико-математических наук, svasilyev@sci.pfu.edu.ru;

Царева Галина Олеговна, аспирант кафедры прикладной информатики и теории вероятностей Российского университета дружбы народов, кандидат физико-математических наук, galinabolotova@gmail.com.